Polymarket Bitcoin rule change cuts Binance settlement spikes
A study found that Binance trading in the final 10 seconds of Polymarket’s five-minute Bitcoin contracts fell to 3.02% after the platform introduced 60-second price averages.
A study by Guangzhe Wang found that Binance trading in the final 10 seconds of Polymarket’s five-minute Bitcoin contract windows fell to 3.02% after the platform introduced 60-second average price benchmarks. The share had been 5.68% under a single-price settlement rule.
Wang examined one-second Binance data from July 1, 2025, through Sept. 4, 2026. The analysis covered 82,752 five-minute boundaries at which only the Polymarket contract settled. Before the five-minute contract launched, trading in the final 10 seconds accounted for 3.05% of Binance volume. The share rose to 5.68% under single-snapshot settlement and later fell to 3.02% under 60-second averaging.
During the first three weeks of the averaged settlement system, the study found no systematic shift into earlier trading intervals. It also found no repeat of a near-strike price reversal pattern identified under the former settlement method.
Polymarket changed its crypto up/down market rules on Aug. 7, replacing single price snapshots with averages based on Chainlink feeds. On Aug. 14 at 00:00 UTC, it extended the benchmark window for five-minute crypto markets to 60 seconds. The contract duration remained five minutes, while the opening and closing prices used to determine the result came from separate 60-second time-weighted average prices.
The Oct. 27 Bitcoin contract covering 12:20 p.m. to 12:25 p.m. Eastern Time listed Chainlink’s BTC/USD 60-second TWAP as its resolution source. The Up outcome paid when the closing benchmark was at least as high as the opening benchmark. Down paid when the closing benchmark was lower.
A 60-second time-weighted average does not necessarily represent 60 equally weighted price observations or the average price during the full five-minute contract. Chainlink reports include a validity period and an observation timestamp, and the time covered by individual reports can vary. The benchmark rolls with the latest observation rather than automatically resetting at the start of each clock minute.
The study measured changes in Binance trading near settlement. It did not assess whether all possible manipulation strategies had ended or calculate traders’ current profits and losses.
An earlier working paper by Dai, Jia and Yu reported spikes in spot-market trading near settlement and subsequent reversals after Polymarket launched its five-minute contracts. An estimate that 821 wallets earned about $8.2 million during earlier pushed cycles relates to that research and the former single-snapshot design.
A profitability assessment under the current rule would need to include both the prediction-market position and any Bitcoin trades used to influence the reference price. Trading costs, execution prices and the effect of those trades on the averaged benchmark would also need to be included.








